Add native OTLP export marker to trace chunks

[mhlidd]

Description

This adds _dd.sdk.otlp_export: "false" to the meta of the first span of each trace chunk. It's set at the start of the is_first_span block in ddtrace_serialize_span_to_rust_span (tracer/serializer.c), right before _dd.tags.process, using the existing ddog_add_str_span_meta_str FFI call. No libdatadog change is needed.

This is part of the "Measuring Adoption of Datadog SDKs in OTel Mode (Traces)" RFC. Today the backend infers which spans came from Datadog SDKs exporting OTLP at ingest time, using telemetry.sdk.name and span_source. That's unreliable: the OpenTelemetry Collector's datadogreceiver hardcodes telemetry.sdk.name:Datadog on native payloads it converts. So tracers now declare their export mode on the wire. "true" goes on the OTLP resource when exporting over OTLP, and "false" goes on the first span of each chunk for native export. An absent marker means an older tracer.

PHP has no OTLP trace export (otlp_traces_export_enabled is hardcoded false), so the value is always "false". The OTLP-only datadog.sdk.semantics resource attribute doesn't apply.

Details:

• Send paths: the sidecar sender, the in-process curl sender and dd_trace_serialize_closed_spans all serialize through this function. No other path writes chunk-level tags.
• Unconditional: the marker isn't gated like process tags, and doesn't guard against a user tag with the same key. That matches the other tracers.
• Dropped spans: is_first_span is true when the trace is still empty, so if stats computation drops a span, the next serialized span gets the marker.

Tests:

• New: tests/ext/sdk_otlp_export_marker.phpt builds two traces of three spans and checks the marker is 'false' on each chunk's first span only.
• Helpers: SpanChecker, SnapshotTestTrait (test-agent ignore list) and dd_dumper.inc ignore the marker the same way they already ignore _dd.tags.process / _dd.svc_src. So existing tag and snapshot assertions don't need to list it.
• Updated expectations: 13 .phpt files under tests/ext/ and 7 under appsec/tests/extension/ that print first-span meta.
• Not run locally: the Docker build environment ran out of disk, so nothing was built or run. CI is the first run. The FFI call matches the existing ddog_add_str_span_meta_str(rust_span, "track_error", "false") usage in the same file. Affected tests were found by grepping, so CI may surface a missed expectation.

Related changes:

• Tracers: Introducing Tags/OTLP Resource Attributes for OTLP Adoption Metrics dd-trace-java#12516, [Tracer] Add OTLP adoption markers to trace payloads dd-trace-dotnet#9310, feat(ddtrace/tracer): add OTLP adoption markers to trace payloads dd-trace-go#5436, feat(tracing): add OTLP adoption markers to trace payloads dd-trace-js#10482, Add native OTLP export marker to trace chunks dd-trace-rb#6390.
• Agent: Propagate _dd.sdk.otlp_export from Tracer to Intake datadog-agent#56749 hoists the marker from the first span of the first chunk into the payload.

This change was written with AI assistance and reviewed before submission.

Reviewer checklist

• Test coverage seems ok.
• Appropriate labels assigned.

🤖 Generated with Claude Code

[datadog-official]

🎉 All green!

🧪 All tests passed
❄️ No new flaky tests detected

🎯 Code Coverage (details)
• Patch Coverage: 100.00%
• Overall Coverage: 68.26% (-0.04%)

_{This comment will be updated automatically if new data arrives.
🔗 Commit SHA: a8f2290 | Docs | View more details | Give us feedback!}

[Copilot]

Copilot review overview

🟢 Approval recommended

The shared serialization path correctly applies the marker once per chunk and the behavior has focused coverage.

Review effort: Balanced
Findings: None

What changed in this PR

Adds the native OTLP export marker to the first serialized span of every trace chunk.

Changes:

• Emits _dd.sdk.otlp_export: "false" during native serialization.
• Adds focused multi-chunk coverage.
• Updates test helpers and affected expectations.

File	Description
tracer/​serializer.c	Adds the marker to each chunk’s first span.
tests/​ext/​sdk_otlp_export_marker.phpt	Verifies marker placement across two chunks.
tests/​ext/​sandbox/​safe_to_string_metadata.phpt	Excludes the new generated marker.
tests/​ext/​sandbox/​dd_dumper.inc	Filters the marker from generic dumps.
tests/​ext/​root_span_url_with_post_no_param.phpt	Updates expected metadata.
tests/​ext/​root_span_url_with_post_no_param_set.phpt	Updates expected metadata.
tests/​ext/​root_span_url_as_resource_names.phpt	Updates expected metadata.
tests/​ext/​root_span_url_as_resource_names_no_host.phpt	Updates expected metadata.
tests/​ext/​referrer_extraction_05.phpt	Updates expected metadata.
tests/​ext/​referrer_extraction_04.phpt	Updates expected metadata.
tests/​ext/​referrer_extraction_03.phpt	Updates expected metadata.
tests/​ext/​referrer_extraction_02.phpt	Updates expected metadata.
tests/​ext/​referrer_extraction_01.phpt	Updates expected metadata.
tests/​ext/​extract_server_values.phpt	Updates expected metadata.
tests/​ext/​distributed_tracing/​distributed_trace_overwrite_active_span.phpt	Filters the generated marker.
tests/​Common/​SpanChecker.php	Ignores the marker unless explicitly asserted.
tests/​Common/​SnapshotTestTrait.php	Excludes the marker from snapshots.
appsec/​tests/​extension/​root_span_add_tag.phpt	Updates expected metadata.
appsec/​tests/​extension/​root_span_add_tag_with_intermediate_spans.phpt	Updates expected metadata.
appsec/​tests/​extension/​rinit_root_span_add_tag.phpt	Updates expected metadata.
appsec/​tests/​extension/​rinit_record_span_tags.phpt	Updates expected metadata.
appsec/​tests/​extension/​rinit_record_span_tags_fail.phpt	Updates expected metadata.
appsec/​tests/​extension/​ddtrace_basic.phpt	Updates expected metadata.
appsec/​tests/​extension/​client_init_record_span_tags.phpt	Updates expected metadata.

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

[bwoebi]

Simple enough, approving it.

Question though: we're in the middle of moving to the v1 protocol in #4046. Will it be a payload-level attribute there?

[mhlidd]

@bwoebi Yup that's what we've done for repos that support V1! Additionally, if OTLP is ever supported in PHP, we would also add it as a Resource Attribute there.